耳鼻咽喉科: オープンアクセス

抽象的な

Acoustic Measures of Phonation during Connected Speech in Adductor Spasmodic Dysphonia

Michael P. Cannito*,Eugene H. Buder,Lesya B. Chorna,Richard Dressler

Objectives: This study examined acoustic measures related to voice production in connected speech of patients with adductor spasmodic dysphonia (ADSD) before and after boltulinum toxin (BT) treatment in comparison with speech of non-dysphonic healthy control (NHC) speakers. Correlations between acoustic measures and perceptual scaling judgments of overall voice quality were also examined.

Methods: Ten patients with moderate-to-severe ADSD were audio-recorded prior to and following BT injection, as were age and gender matched NHCs. Signal processing algorithms were employed to extract fundamental frequency (f 0 ) and intensity (dB) of speakers’ digitized oral readings. Control procedures minimized the influence of severely aperiodic phonation and obstruent consonant production on the analysis of modal f 0 and related variables such as signal-to-noise ratio (SNR). Measures of severe aperiodicity, with f 0 < 80 Hz, were compared with modal f 0 data. Perceptual judgments of voice quality were obtained from expert voice clinicians under rigorously controlled listening conditions.

Results: Acoustic measures of number of low frequency f 0 events, coefficient of variation of modal f 0 , and SNR demonstrated statistically significant differences for ADSD speech before and after treatment, and differentiated between ADSD and NHC speakers. Mean modal f 0 did not differentiate among speaker conditions. All other measures differentiated untreated ADSD speech from that of NHCs; however, only selected measures demonstrated differences between NHC and ADSD speakers following BT injection. Coefficient of variation of modal f 0 and SNR were moderately correlated with expert listener judgments of voice quality. Conclusion: Acoustic measures of connected speech derived from carefully edited modal f 0 tracks and intensity contours were effective for characterizing ADSD speech before and after treatment, and for differentiating it from that of NHCs. Acoustic measures were highly reliable and significantly related to voice quality scaling by expert listeners.